This almost worked <g>.

DECLARE @tableName NVARCHAR(50);
DECLARE @TemplateSQL NVARCHAR(500)=
'SELECT *
INTO #tempData
FROM dbo.MikeyData n
EXCEPT
SELECT *
FROM #tempData d;';

DECLARE @InsertSQL NVARCHAR(500);

DECLARE tables_cursor CURSOR  FAST_FORWARD
    FOR 
SELECT ao.name
FROM...

What are your options?

If you can use SSRS, it's stupid easy.

Can you use a matrix in .NET? (Well, SSRS is an ASP.net application, so I would think so.)

I created a table for this... (Far easier to do in SSRS with a matrix visual)

now someone who's better at Pivot than I am can have a play.

            
September 1, 2020 at 2:26 pm
            
#3785215

            
            
                                                
        

Got some sample records?  And how about a couple of CREATE TABLE scripts. Or do you have something against tested queries?

So what's your question? You want to split the last and first names into separate columns in your SQL query? (I would do that... makes querying much easier and faster).

Could you explain the logic behind how your desired ranking  works?

I rewrote your create table and inserts, because they didn't run. You don't need DateTime2 if you're not using time.

CREATE TABLE [test2]
(
[InDate] [date] NULL,
[OutDate] [date] NULL,
[weight] [float]...

• This reply was modified 5 years, 8 months ago by pietlinden.

Not for me. Can you show at least a partial expected result from the query using the data you provided?

This worked:

SELECT t.*
, ca.value
FROM #tblTest t
CROSS APPLY STRING_SPLIT(t.FileData, ';') ca
WHERE LEN(TRIM(ca.Value))>0;

Use ISBLANK([ColumnWithNulls],"AlternativeValue")

ISBLANK is analogous to T-SQL ISNULL()

Is there a really short article showing how to do this from Python outside of SQL Server? (Or from Jupyter Notebooks?)

Then you can use pandas to do the data analysis...

FILTER(<tableObject>, ISNONBLANK([Column]) )?

(Maybe it made sense to me because I used to work on "databases" where nobody understood normalization... except my databases were small (in terms of records)... there were stupid wide...

That's what I assumed it meant. Other interpretations didn't make sense. I suppose writing the table name and column name to at least a temporary table and then using that...

I kinda hope there's a better way of doing this... but this works to create a SQL statement containing only the columns with data (in any of the rows)...

            
August 25, 2020 at 6:03 am
            
#3783170